TheAncientGeek comments on Is Global Reinforcement Learning (RL) a Fantasy?

[TheAncientGeek]

If we know nothing about architecture X, surely we should adopt a Bayesian 50:50 about whether some other architecture is applicable to it.